This is another installment of Mighty AI's "Conversations in Machine Learning" blog series. Each week, our content human, Cassie, shares a summary of a recent conversation we had with a machine learning team and potential customer--what they're building, how they're handling training data today, etc. Read more about the series here. Each of those would be appropriate depending on the context--you'd connect its brand name to a search engine (because it is one, among other things), you'd call it a technology company with a glance-over of its corporate site, and you'd say it's a machine learning company if you dug into its products. That's because these days, machine learning is powering virtually everything the company does and is behind nearly everything it builds. To back up a bit, this is a European technology firm with deep roots in search and respectably deep roots in machine learning, too (employees there developed a proprietary ML methodology nearly a decade ago, and creating a whole new methodology sounds impressive to me at least).

Across enterprises, ever-expanding stores of data remain unstructured and unanalyzed. Few organizations have been able to explore nontraditional data sources such as image, audio, and video files; the torrent of machine and sensor information generated by the Internet of Things; and the enormous troves of raw data found in the unexplored recesses of the "deep web." However, recent advances in computer vision, pattern recognition, and cognitive analytics are making it possible for companies to shine a light on these untapped sources and derive insights that lead to better experiences and decision making across the business. In this age of technology-driven enlightenment, data is our competitive currency. Buried within raw information generated in mind-boggling volumes by transactional systems, social media, search engines, and countless other technologies are critical strategic, customer, and operational insights that, once illuminated by analytics, can validate or clarify assumptions, inform decision making, and help chart new paths to the future. Until recently, taking a passive, backward-looking approach to data and analytics was standard practice. With the ultimate goal of "generating a report," organizations frequently applied analytics capabilities to limited samples of structured data siloed within a specific system or company function. Moreover, nagging quality issues with master data, lack of user sophistication, and the inability to bring together data from across enterprise systems often colluded to produce insights that were at best limited in scope and, at worst, misleading.

Local Process Models (LPM) describe structured fragments of process behavior occurring in the context of less structured business processes. Traditional LPM discovery aims to generate a collection of process models that describe highly frequent behavior, but these models do not always provide useful answers for questions posed by process analysts aiming at business process improvement. We propose a framework for goal-driven LPM discovery, based on utility functions and constraints. We describe four scopes on which these utility functions and constrains can be defined, and show that utility functions and constraints on different scopes can be combined to form composite utility functions/constraints. Finally, we demonstrate the applicability of our approach by presenting several actionable business insights discovered with LPM discovery on two real life data sets.

Google just killed the Captcha, perhaps the most obstructive thing on the entire internet. For years, Captcha served as the primary way of telling humans and robots apart on the internet. It made sure that the person looking to access a website was actually a human being – ensuring that robots couldn't be used to send spam or flood a website with requests, for instance. But over time, robots have gradually become too clever for the often simple tests – which early on required people to transcribe hard-to-read text. With that, the technologies have become more complex, too.

As the artificial intelligence market is projected to grow by 53% in by 2020, advertisers are looking for ways to use the technology to their advantage. Vernon Vasu, CMO at ReFUEL4 states that researchers are looking into using AI for creative development in the future, but for now advertisers can use AI's incredible data mining and organizing capabilities to understand audiences like never before Artificial intelligence is one of the most buzzed-about terms in technology. The AI market is estimated to reach $5.05 billion USD by 2020, up from $419.7 million USD in 2014 – a 53% increase. With the launch of Facebook's chatbots, Amazon's Echo, and IBM's Watson, companies in many fields are considering how they can use new AI tools to their advantage. Advertising agencies that use AI, machine learning, and image recognition are hyper-targeting consumers by learning their interests and tastes.

Phillip De Leon has been named associate dean of research and doctoral studies for the New Mexico State University College of Engineering. LAS CRUCES -- Through the use of New Mexico State University's High Performance Computing system, a supercomputer known as Joker, Phillip De Leon, associate dean for research in the College of Engineering, not only conducted research but students in his graduate electrical engineering course also used the system. In the Pattern Recognition and Machine Learning course, which is a data science class De Leon taught in the fall, graduate students used Joker on projects that included developing machine learning codes and evaluating the models with standard datasets. "These projects including identifying a song much like the Shazam app, recognizing handwritten digits like ZIP codes, classifying email as ham or spam, analyzing Twitter feeds, etc.," De Leon said. "Being able to use this system allowed the students to experiment and tune their codes much faster since everything ran much faster. It also allowed for big datasets to be used in training and evaluation."

Salesforce is getting into the computer vision business with a new tool designed to let users easily train a custom image recognition system. Einstein Vision, as it's known, allows users to upload sets of images and classify them in a series of categories. After that, the system will create a recognizer based on machine learning technology that will identify future images fed into it. While Salesforce customers will have to wait a couple weeks before Vision is generally available, the company announced Tuesday that other Einstein features based on machine learning techniques are live. It's the latest step in a long journey for Salesforce, which began touting Einstein last year and demoed those capabilities at its Dreamforce conference. The company is facing heavy competition, and Einstein might give it an edge against the likes of Microsoft and Oracle.

Facebook is using a combination of pattern recognition, live chat support from crisis support organizations and other tools to prevent suicide, with a focus on its Live service. There is one death by suicide every 40 seconds and over 800,000 people kill themselves every year, according to the World Health Organization. "Facebook is in a unique position--through friendships on the site--to help connect a person in distress with people who can support them," the company said Wednesday. The move by Facebook appears to aim to prevent the live-streaming of suicides on the Live platform, which was launched in April last year, and allows people, public figures and pages to share live videos with friends and followers. The company said that its suicide prevention tools for Facebook posts will now be integrated into Live, giving people watching a live video the option to reach out to the person directly and to report the video to the company.

Taking a souvenir home from an art gallery no longer has to mean a trip to the gift shop. A new app lets people scan a work of art with their smartphone camera to find out more about it and save a digital copy. The app, called Smartify, uses image recognition to identify scanned artworks and provide people with additional information about them. Users can then add the works to their own digital collection. Smartify co-founder Thanos Kokkiniotis describes it as a combination of the music discovery service Spotify and music recognition app Shazam – but for visual works.

Facebook is trialling new tools to help with suicide prevention efforts. One approach will use artificial intelligence to identify concerning posts and make it easier for other people to report them. Facebook says it will use pattern recognition algorithms to spot posts that could indicate someone is suicidal and help their friends to flag this content by making the option to report posts about "suicide and self injury" more prominent for those that are considered potentially concerning. The algorithms are trained on posts that have previously been reported. It will also use pattern recognition to flag posts "very likely to include thoughts of suicide" so that its community operations team can take action even if the post is not reported.